Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A method comprising: detecting and extracting face data corresponding to at least a first candidate face of a first image and a second candidate face of a second image, the first image is associated with a first light intensity and the second image is associated with a second light intensity that is different from the first light intensity; analyzing the face data to determine whether the first candidate face corresponds to a first area in the first image that is substantially the same as a second area of the second candidate face in the second image; evaluating, via a processor, data associated with the first and second areas to determine whether the first and second candidate faces are valid faces or invalid faces of at least one individual; determining that the first area and second areas are the same in an instance in which a location of the first area is within a predetermined threshold of a corresponding location of the second area, wherein the predetermined threshold is equal to zero when images with varying light intensities, provided by an image capturing device, are captured with no movement between the image capturing device, and the first and second candidate faces; in an instance when the first and second candidate faces are valid, enabling conversion of the extracted face data of the first and second images into one or more facial features to obtain corresponding images of the first and second candidate faces; normalizing the converted facial features to obtain at least one representation associated with an image of a face corresponding to a person; and analyzing data associated with the at least one representation and data uniquely identifying the person based upon the normalized converted facial features associated with the at least one representation to recognize the one or more corresponding faces.
A method for facial recognition handles varying lighting conditions by: 1) Detecting face candidates in two images captured with different light intensities (e.g., with and without flash). 2) Checking if the face candidates represent the same area in both images. The areas are considered the same if their locations are within a threshold (zero if the camera is still). 3) Validating each face candidate based on the area data. 4) If both are valid, the method converts extracted face data into facial features. 5) It normalizes these features to create a unified representation of the face. 6) Finally, it analyzes this representation against known identities to recognize the face.
2. The method of claim 1 , further comprising: determining that the first and second candidate faces are valid faces in response to determining that the first area in the first image corresponds to substantially the same location of the second area in the second image.
The facial recognition method described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, and analyzes against known identities, further determines that the face candidates are valid ONLY if they appear in approximately the same location within both images. If the locations don't correspond, the faces are considered invalid.
3. The method of claim 1 , further comprising: determining that the first and second candidate faces are invalid faces in response to determining that the first area in the first image does not correspond to substantially the same location of the second area in the second image.
This invention relates to face detection and validation in image processing, specifically addressing the challenge of accurately identifying and verifying faces across multiple images. The method involves analyzing two images to determine whether candidate faces detected in each image correspond to the same physical location. If the detected face regions in the two images do not align with the same spatial location, the faces are deemed invalid. This process helps eliminate false positives and ensures that only valid, consistent face detections are retained. The method may also include comparing additional features, such as facial landmarks or image metadata, to further validate the detected faces. The invention is particularly useful in applications like surveillance, biometric authentication, and facial recognition systems where accurate face detection is critical. By cross-referencing face locations between images, the system improves reliability and reduces errors in face identification tasks. The technique can be applied in real-time systems or batch processing workflows, enhancing the robustness of face detection algorithms in various imaging scenarios.
4. The method of claim 1 wherein: the second light intensity is generated based in part on a flash of light provided by the imaging capture device and the first light intensity is generated without the flash.
The facial recognition method described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, and analyzes against known identities, uses images where the different light intensities are achieved by using a camera flash for one image and not using the flash for the other.
5. The method of claim 3 , further comprising: enabling removal of the first and second candidate faces from consideration in response to determining that the first and second candidate faces are invalid.
The facial recognition method described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, analyzes against known identities, and determines invalid faces based on mismatched area locations, further removes these invalid face candidates from any further processing steps, such as attempting to identify the individual.
6. The method of claim 3 , further comprising: enabling provision of display of visible indicia denoting that the first and second candidate faces correspond to detection of false faces in response to determining that the first and second candidate faces correspond to the invalid faces.
The facial recognition method described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, analyzes against known identities, and determines invalid faces based on mismatched area locations, further displays a visual indicator when it identifies invalid face candidates, alerting the user that a false face detection has occurred.
7. The method of claim 1 , further comprising: detecting whether another determined representation associated with another face of at least one of the images corresponds to the representation; and determining that the another face corresponds to the face of the person in an instance in which the determined representation corresponds to the representation.
The facial recognition method described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, and analyzes against known identities, further improves recognition by detecting additional faces in the images and comparing their representations against the original representation. If a match is found, it confirms that the additional face also belongs to the same person.
8. An apparatus comprising: at least one processor; and at least one memory including computer program code, the at least one memory and the computer program code configured to, with the at least one processor, cause the apparatus to perform at least the following: detect and extract face data corresponding to at least a first candidate face of a first image and a second candidate face of a second image, the first image is associated with a first light intensity and the second image is associated with a second light intensity that is different from the first light intensity; analyze the face data to determine whether the first candidate face corresponds to a first area in the first image that is substantially the same as a second area of the second candidate face in the second image; evaluate data associated with the first and second areas to determine whether the first and second candidate faces are valid faces or invalid faces of at least one individual; determine that the first area and the second area are the same in an instance in which a location of the first area is within a predetermined threshold of a corresponding location of the second area, wherein the predetermined threshold is equal to zero when images with varying light intensities, provided by an image capturing device, are captured with no movement between the image capturing device, and the first and second candidate faces; in an instance when the first and second candidate faces are valid, enable conversion of the extracted face data of the first and second images into one or more facial features to obtain corresponding images of the first and second candidate faces; normalize the converted facial features to obtain at least one representation associated with an image of a face corresponding to a person; and analyze data associated with the at least one representation and data uniquely identifying the person based upon the normalized converted facial features associated with the at least one representation to recognize the one or more corresponding faces.
An apparatus (e.g., a device with a processor and memory) performs facial recognition by: 1) Detecting face candidates in two images captured with different light intensities (e.g., with and without flash). 2) Checking if the face candidates represent the same area in both images. The areas are considered the same if their locations are within a threshold (zero if the camera is still). 3) Validating each face candidate based on the area data. 4) If both are valid, converting extracted face data into facial features. 5) Normalizing these features to create a unified representation of the face. 6) Analyzing this representation against known identities to recognize the face.
9. The apparatus of claim 8 , wherein the at least one memory and the computer program code are further configured to, with the processor, cause the apparatus to: determine that the first and second candidate faces are valid faces in response to determining that the first area in the first image corresponds to substantially the same location of the second area in the second image.
The facial recognition apparatus described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, and analyzes against known identities, further determines that the face candidates are valid ONLY if they appear in approximately the same location within both images.
10. The apparatus of claim 8 , wherein the at least one memory and the computer program code are further configured to, with the processor, cause the apparatus to: determine that the first and second candidate faces are invalid faces in response to determining that the first area in the first image does not correspond to substantially the same location of the second area in the second image.
The facial recognition apparatus described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, and analyzes against known identities, further determines that the face candidates are invalid IF the detected areas in the images do NOT correspond to the same location.
11. The apparatus of claim 8 wherein the second light intensity is generated based in part on a flash of light provided by the imaging capture device and the first light intensity is generated without the flash.
The facial recognition apparatus described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, and analyzes against known identities, uses images where the different light intensities are achieved by using a camera flash for one image and not using the flash for the other.
12. The apparatus of claim 10 , wherein the at least one memory and the computer program code are further configured to, with the processor, cause the apparatus to: enable removal of the first and second candidate faces from consideration in response to determining that the first and second candidate faces are invalid.
The facial recognition apparatus described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, analyzes against known identities, and determines invalid faces based on mismatched area locations, further removes these invalid face candidates from any further processing steps, such as attempting to identify the individual.
13. The apparatus of claim 10 , wherein the at least one memory and the computer program code are further configured to, with the processor, cause the apparatus to: enable provision of display of visible indicia denoting that the first and second candidate faces correspond to detection of false faces in response to determining that the first and second candidate faces correspond to the invalid faces.
The facial recognition apparatus described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, analyzes against known identities, and determines invalid faces based on mismatched area locations, further displays a visual indicator when it identifies invalid face candidates, alerting the user that a false face detection has occurred.
14. The apparatus of claim 8 , wherein the at least one memory and the computer program code are further configured to, with the processor, cause the apparatus to: detect whether another determined representation associated with another face of at least one of the images corresponds to the representation; and determine that the another face corresponds to the face of the person in an instance in which the determined representation corresponds to the representation.
The facial recognition apparatus described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, and analyzes against known identities, further improves recognition by detecting additional faces in the images and comparing their representations against the original representation. If a match is found, it confirms that the additional face also belongs to the same person.
15. A computer program product comprising at least one non-transitory computer-readable storage medium having computer-executable program code instructions stored therein, the computer-executable program code instructions configured to: enable detection and extraction of face data corresponding to at least a first candidate face of a first image and a second candidate face of a second image, the first image is associated with a first light intensity and the second image is associated with a second light intensity that is different from the first light intensity; analyze the face data to determine whether the first candidate face corresponds to a first area in the first image that is substantially the same as a second area of the second candidate face in the second image; evaluate data associated with the first and second areas to determine whether the first and second candidate faces are valid faces or invalid faces of at least one individual; determine that the first area and second areas are the same in an instance in which a location of the first area is within a predetermined threshold of a corresponding location of the second area, wherein the predetermined threshold is equal to zero when images with varying light intensities, provided by an image capturing device, are captured with no movement between the image capturing device, and the first and second candidate faces; in an instance when the first and second candidate faces are valid, enable conversion of the extracted face data of the first and second images into one or more facial features to obtain corresponding images of the first and second candidate faces; normalize the converted facial features to obtain at least one representation associated with an image of a face corresponding to a person; and analyze data associated with the at least one representation and data uniquely identifying the person based upon the normalized converted facial features associated with the at least one representation to recognize the one or more corresponding faces.
A computer program stored on a non-transitory medium performs facial recognition by: 1) Detecting face candidates in two images captured with different light intensities (e.g., with and without flash). 2) Checking if the face candidates represent the same area in both images. The areas are considered the same if their locations are within a threshold (zero if the camera is still). 3) Validating each face candidate based on the area data. 4) If both are valid, converting extracted face data into facial features. 5) Normalizing these features to create a unified representation of the face. 6) Analyzing this representation against known identities to recognize the face.
16. The computer program product of claim 15 , further comprising program code instructions configured to: determine that the first and second candidate faces are valid faces in response to determining that the first area in the first image corresponds to substantially the same location of the second area in the second image.
The computer program for facial recognition described previously, which detects face candidates in two images with different lighting, checks for matching areas, validates faces, converts data to facial features, normalizes features, and analyzes against known identities, further determines that the face candidates are valid ONLY if they appear in approximately the same location within both images.
17. The method of claim 7 , further comprising: enabling storage of the another determined representation associated with another face of at least one of the newly captured images in an instance in which the another face is valid.
The facial recognition method described previously, which detects additional faces in the images and compares their representations against the original representation and improves recognition, further stores the representation of an additional face ONLY if it is determined to be a valid face.
18. The apparatus of claim 14 , wherein the at least one memory and the computer program code are further configured to, with the processor, cause the apparatus to: enable storage of the another determined representation associated with another face of at least one of the newly captured images in an instance in which the another face is valid.
The facial recognition apparatus described previously, which detects additional faces in the images and compares their representations against the original representation and improves recognition, further stores the representation of an additional face ONLY if it is determined to be a valid face.
Unknown
September 12, 2017
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.